Drop in cancer screening during COVID-19 may aid research on overdiagnosis

An article published on February 11 in the journal STAT tells us about a study project that consists of examining a "natural experiment" to evaluate data such as mortality or overdiagnosis in cancer screening. 

This natural experiment concerns the decline in screening tests during the pandemic. We could then examine and gauge the overdiagnosis of cancers, the real burden of screening, since it throws healthy people into the hell of the disease.

This examination would thus be done at the end of the pandemic, which seems more ethical than a prospective study where patients would have to be randomized in order to be subjected to screening or not.

(Editor's note: however, such a study seems to us to be quite feasible with good information for participants, but more costly and more complicated to set up[1]. It can just be considered that conducting a randomized study on overdiagnosis with two groups of people, perfectly informed before participation in the study, is certainly more ethical than subjecting entire populations, as is currently the case, to highly controversial screenings as to their effectiveness, without any fair information on the benefit-risk balance. This is the case for prostate cancer screening (many men are prescribed a PSA dosage without any explanation) and for breast cancer (many women, uninformed, believe that breast cancer screening is mandatory and will save their lives)).

The author of the article, Ms. E. Cooney, is a general assignment journalist at STAT, a journal of biotechnology, pharmaceutical, policy and life sciences analysis, which she joined in 2017. Previously, she was a blogger at the Boston Globe, before moving to the science editor at the Broad Institute of MIT and Harvard (a biomedical and genomics research center located in Cambridge, Massachusetts, U.S.A.).


Ned Sharpless, current director of the U.S. National Cancer Institute (NCI) and professor of medicine and genetics, was alarmed by the sharp drop in the number of screenings by colonoscopies, mammograms and other cancer screening tests. His concern was motivated by models predicting an explosion in cancer rates if screening was not performed. (In France, we also had our national cancer centres forecasting the worst consequences in the event of non-screening[2]).

However, it will be the delays and postponements in the arrival of patients that will have the most consequences, according to the Grouvid study[3] presented by the statistician Aurélie Bardet of the Gustave-Roussy Institute in Villejuif (Val-de-Marne).

But in January of this year, Mr. Sharpless questioned the downside of early detection: overdiagnosis, when asymptomatic cancers that may not develop and harm the patient are detected, and the overtreatment that accompanies it. The pandemic, he says, could be an opportunity to resolve a long-standing controversy over the extent to which the disadvantages of cancer screening outweigh its benefits. "Knowing the extent to which overdiagnosis and overtreatment actually occur during cancer screening is a very complicated subject," he said. For him, the pandemic has provided an interesting natural experiment, where we could examine some of these tumours, diagnosed later. Is the fate of patients really less favorable because of a later diagnosis of their tumor?

If delays in screening - depending on the cancer and the screening test - do not lead to worse outcomes for most patients, then they could provide valuable information when we will emerge from the pandemic. To demonstrate that non-use of screening is not detrimental, the decline in screening rates should lead to a decrease in overdiagnosis, and not concomitantly result in a significant increase in the number of patients developing disabling or fatal cancers.

(Editor's note: Caution, raw rate data should be examined. In fact, if overdiagnosis could be completely eliminated, the part, i.e., the proportion of serious cancers, would then appear to be greater in the total number of cancers minus overdiagnoses, which usually amplify the total number of cancers. In fact, the proportion of serious cancers is diluted in the total cancer figure when the contribution of overdiagnosis is included in the total cancer figure. 

It is therefore necessary to look at the crude rate and not at the percentage of serious cancers in the total number of cancers identified).

For Mr. Sharpless, the time has come to examine more closely the natural history of cancers during the postponement period, and overtreatments.

Observations already available

For Clifford Hudis, a breast cancer specialist and CEO of the American Society of Clinical Oncology, tests such as colonoscopy are clearly useful. But for other cancers, the impact on survival has been less obvious.
Mr. Hudis emphasizes the difference between screening a person with no symptoms and diagnosing a person with clinical signs. ( It is obvious that under no circumstances consultation should be postponed in case of a clinical symptom).
For him, the evidence is irrefutable: Pap smears and HPV tests to detect precancerous cells in the cervix have significantly reduced mortality rates. Colonoscopy and other screening tests for colorectal cancer in adults have been similarly successful, to the point where an expert panel is recommending screening for colorectal cancer in younger people.
(Editor's note: this opinion is much more nuanced according to experts and studies [4] Perhaps we are still lacking hindsight).

There is much less certainty about the effectiveness of mammography and CT screening for lung cancer, while the use of PSA screening for prostate cancer continues to be controversial[5]. Yet we have "natural experience" with this screening.
Routine PSA testing opened the door to a sharp increase in the incidence of prostate cancer diagnoses, which rose by about 16% per year from 1988 to 1992, then by 9% per year until it stabilized in the late 1990s.
After the U.S. Preventive Services Task Force (USPSTF) changed its recommendation in 2012 to advise men against routine PSA testing, the incidence of prostate cancer has finally stabilized, in contrast to what was happening before widespread testing. And mortality has not changed!

Cooney also relates the position of Barnett Kramer, former director of the NCI's Division of Cancer Prevention: "It's not just prostate cancer where over-diagnosis and over-treatment are a concern," he says. « There are other slow-growing cancers that would never cause suffering during a patient's natural lifetime, and there are also cancers that never progress ».
“You introduce a screening test, in this case for thyroid cancer, kidney cancer, and melanoma, and you dramatically increase the incidence of cancer and prevalence of those cancers,” he declared in front of an association of health journalists.
“There are more and more people that are diagnosed with cancers, and yet you make very little impact on the mortality. They are cured. But they never would have gone on to die of the cancer anyway.”

For instance, while Mr. Kramer sees real harm in overdiagnosis, he does not want this message to encourage people to avoid seeking medical care for symptomatic cancer. “People should be alerted to making sure they seek medical attention at the earliest signs of symptoms,” he said. « We do know that ignoring advancing symptomatic disease is harmful.”

The author also quotes Otis Brawley, former Medical and Scientific Director of the American Cancer Society and now a professor at Johns Hopkins University: “There are cancers that don’t need to be cured. And that’s at least 60% of all prostate cancers and maybe 20% of breast cancers don’t need to be cured.”
(Editor's note: the 20% estimate concerns invasive carcinoma according to the first randomized studies on Canadian trials for example [8], but on the one hand these estimates are now being revised upwards, and on the other hand, if we add carcinomas in situ, 80% of which would not need to be treated, the over-diagnosis of breast cancer is estimated at almost 50%, which means that it could concern one cancer in two detected by mammographic screening [9]).

According to Dr. Sharpless, most cancers are discovered following the appearance of symptoms.
Nevertheless for him “there are plenty of people alive today because they had an asymptomatic lesion detected on some screening exam that was removed, resected, treated, and they’re cured of what would’ve been a very symptomatic cancer. The opposite argument is true, too.”, meaning: many of the living people treated by removing, resecting lesions that would never have become symptomatic.

Another factor to consider is that cancer therapies have improved. Some may be diagnosed at a later stage with delayed screening, but they are curable with effective therapies," says Sharpless. So the higher grades can be apprehended by new therapeutic approaches.

(Editor's note: P. Autier's study noted: The influence of mammography screening on mortality declines with the increasing effectiveness of cancer therapies.)

Which method for evaluation?

Eric Feuer, founder and leader of the NCI Cancer Surveillance Modeling Network Evaluation Project, has worked on NCI models predicting excess mortality from breast and colorectal cancer due to delayed screening. 
He stated that the widespread use of the PSA test was also a natural experiment.
“When screening goes up rapidly, you’re taking cases from the future,” Feuer said. « Some of those cases never would have caused symptoms, but they made incidence rise. The problem is that the PSA test doesn’t accurately predict which cancers, with or without symptoms, will be harmful and which won’t. »

For breast and colorectal cancer, Dr. Feuer will examine data from the Surveillance, Epidemiology, and End Results program (SEER) and other NCI data to see if the decline in screening rates has been followed by a decline and then an increase in incidence. He will monitor rates of positive screening tests, the stage of confirmed cancer diagnosis, and mortality data.
In addition to the SEER, the Population-based Research to Optimize the Screening Process (PROSPR) observational database will also be used to discern the effects that decreased cancer screening may have on the stage at which cancer is diagnosed, among other measures.

All this is supposed to allow the estimation of overdiagnosis, if the data return is powerful enough.

Mr. Kramer (former director of the NCI's Division of Cancer Prevention) has requested that PSA be removed from his usual blood test panel.

Professor Brawley (Johns Hopkins University professor) thinks Covid-19 is the imminent danger we should consider now. His hospital has cancelled elective operations, including radical prostatectomies.

The natural experiment on screening tests will take some time to show results.

“We’ll know in 10 years,” NCI’s Feuer said.


The director of the Canadian Breast Imaging Society said her group already has a study underway. The full transcript of the interview can be found at this link: https://www.cbc.ca/radio/thecurrent/the-current-for-feb-9-2021-1.5906730/february-9-2021-episode-transcript-1.5907645

MATT GALLOWAY: Tell me more about that. I mean, in the six months that this pandemic would have perhaps derailed or slowed down or stopped screening, what would be the change in that in terms of, you know, where cancer might be, but also, as you say, survival rates?

JEAN SEELY: Well, it's too early for us to know. And we're starting a study to look at this because we believe that this is a factor that's happening across the country. There was a modelling study done by the group at Sunnybrook, and they used a mathematical model called OncoSim, where they estimated that a six-month delay in screening would lead to 670 more advanced breast cancers in Canada and 250 more breast cancer deaths in the next 10 years. And delaying that even more than six months, which may happen with people stopping or naturally not referring themselves, would lead to an even higher number of deaths. So we are seeing this across the country.

So we can easily imagine how breast medical imaging experts will design the research to their advantage, and how they will promote it. The debate will never end, and a battle of "models" will be engaged.

Opinion of Dr. Vincent Robert, our medical statistician :

"Apart from the fact that this is only a model (and with another model we would certainly find something else), which statistician, and with which tool, would be able to spot an increase of 250 deaths among the random fluctuations of the annual number of deaths (with 12000 deaths per year on average, the confidence interval of the annual number of deaths in France has an amplitude of about 500 deaths, i.e. the annual number of deaths naturally fluctuates by much more than 250; or, if you prefer, an increase of 250 deaths will not appear to be statistically significant). "

To be continued..…


[1] https://cancer-rose.fr/my-pebs/methodologie/ - choose article: "l'étude dont on rêvait"

[2] https://cancer-rose.fr/2020/10/06/langoisse-des-thuriferaires-du-rose-face-a-la-decroissance-de-participation/

[3] https://cancer-rose.fr/en/2021/01/18/covid-19-pandemic-and-cancer-management/

[4] https://bmjopen.bmj.com/content/9/10/e032773

[5] https://cancer-rose.fr/en/2021/02/11/parallel-to-breast-screening-prostate-screening-overdiagnosis-as-well/

[6] https://cancer-rose.fr/en/2021/02/01/overdiagnosis-of-thyroid-cancer-another-womans-concern/

[7] https://www.youtube.com/watch?v=068KMIe-gys&feature=emb_logo
Dr. Adewole Adamson's observation is frightening: no reduction in mortality and massive overdiagnosis due to the fact that the tolerance thresholds of dermatologists and anatomo-pathologists are lowered in front of skin lesions.

Dermatologists are increasingly and rapidly asking for biopsies, while anatomical pathologists prefer to upgrade their diagnosis of lesions examined under the microscope (i.e., when in doubt, classify as malignant lesions that are simply dubious and that could only be monitored), giving rise to an apparent melanoma epidemic with even more artificial "survivors".The vicious circle is endless, prompting patients and doctors to do more and more routine skin examinations.

[8] https://cancer-rose.fr/en/2021/01/06/miller-study/

[9] https://cancer-rose.fr/en/2020/12/17/mammography-screening-a-major-issue-in-medicine/

Cancer Rose est un collectif de professionnels de la santé, rassemblés en association. Cancer Rose fonctionne sans publicité, sans conflit d’intérêt, sans subvention. Merci de soutenir notre action sur HelloAsso.

Cancer Rose is a French non-profit organization of health care professionals. Cancer Rose performs its activity without advertising, conflict of interest, subsidies. Thank you to support our activity on HelloAsso.